Measuring Technological Change through an Extended Structural Decomposition Analysis: An Application to EU-28 Primary Sectors (2010–2015)

Round 1

Reviewer 1 Report

My review comments to the manuscript are written in the attached file.  

Comments for author File: Comments.pdf

Author Response

P1L24:

A comment to that respect is introduced. Under our opinion, reference done to Miller & Blair (2009) is adequate.

P1L38:

Firstly, different from ours, Wood & Lenzen (2009) inverse decomposition is based on the classical . In our view there is not a relevant relationship with the partition proposed by Defourny & Thorbecke (1984).

P3L111:

A brief comment in included.

P5L140:

In standard input-output notation the generic element of the Leontief´s inverse is noted as α_ij (see Miller & Blair, 2009 or Economic System Research guidelines).

Equation (4):

We admit that the introduction of new notation does not facilitate the normalization´s description. It might be better to use diag(A), diag(A²), ...

Equation (7):

The α_ij elements are defined in (3), distinguishing if the entries are in the main diagonal of the matrix (i=j) or not.

P11L267:

The present work is devoted to present an alternative methodology. Since this new method can be used to analyse all the sectors of an input-output table a change in title has been decided. In order to explain how the proposal functions, only the first sectors of a table were considered.

In the annexed file the calculations for the remaining industries are presented. As it would be difficult to graphically show results for all industries, we have focused just in the first ones. Thus, these are the ones that appear in the tables displayed in the article.

We propose to publish the present investigation alongside a supplementary data file.

P11L284:

A relevant relation with this article´s proposal was not found.

Minor comments:

Our use of English has been revised. We hope all mistakes have been corrected.

Reviewer 2 Report

Based on IOA and SDA, this study determined the causes of changes in production in agriculture, forestry and fishing sectors in six EU-28 countries in the 2010-2015 period. Technically, this study focused on disaggregation technological change by distribution factors associated with a specific normalisation of the Leontief inverse. I think it is a very innovative contribution to existing research. However, there is still some room for modification.

1. In introduction section, the authors list a mount of works in the field of agricultural operations. However, a comment is needed at the end of this paragraph, especially about the knowledge gap in this field.
2. This paper focused on the agriculture, forestry and fishing sectors in six EU-28 countries. Please tell us why the authors choose these six countries as a case study. Do the agriculture, forestry and fishing sectors in these countries have some special characteristics compared with other countries ?
3. The innovations and contributions of this study should be further  highlighted, especially for the application of methodology.
4. In the conclusions section, the authors should put forward the visions of future work. Whether the methodology proposed by the authors can be expanded to other sectors in other countries to reveal some more interesting mechanism.

Author Response

As for comment 1. A paragraph is added in accordance with this recommendation.

As for the rest of comments. A change on the title has been decided. This way, we want to focus on the methodological proposal rather than the specific sectors included in its empirical application. In the attached file it can be seen how calculations were done for all industries. It would be difficult to graphically show all results at the same time.

Reviewer 3 Report

This paper focus on an interesting topic but the innovation that presents is scarce. For example, it says that the decomposition of Leontief matrix has not been done and that is not entirely correct because in last years works from Dietzenbacher and Lahr have showed how to decompose parts of the L matrix. 

 

The authors choose a very particular set of countries and years. This reduces the interest of the paper. It is not clear why they do it. They say that it is because there was no price changes. Well, they could have transformed the IO tables to constant prices as suggested by Lan, Malik and Lenzen et al. (2016). In fact, using current tables may produced biased results.  

It is not clear for me why they use the normalisation of the Leontief tables. They just say they do but they do not explain why. It is not clear if it was because they were working with different tables. 

 

The idea to decompose the output is interesting but not new. However, this decomposition in three sectors is particularly limited. Yet, it is important to underline that it is well done. However, papers decomposition the output in 5,6 or 7 factors have already been published and so, the methodology seems very limited. 

 

The interpretation and implications of the results should be better stated. It is not clear for me why this research is important for those studying agriculture or forestry studies? If you select this sectors then you must adequate and confront your results at the light of what are the other studies in these areas (even if they use distinct methodologies). 

 

Also, you should clearly state that your work is limited because further decompositions of final demand were possible of being done. Moreover, you could have applied this technique to GVA and employment and understand if the same behavior was identified. 

Author Response

Comment 1.

Firstly, it is to be noted that the expression related with the inverse´s decomposition has been moderated. SDA literature normally focuses on final demand factors. When it comes to decompose technological change, the standard Taylor series (I+A+A² +...) is used. Hence, the methodology here proposed could constitute some novelty to be considered in these contexts.

This methodology is different from the one exposed by Dietzenbacher & Lahr. A recent article published in the Journal of Economic Structures also addresses how to decompose Leontief´s inverse. It has been included as a reference. However, theses approaches are different from the one in the present investigation.

Comment 2.

The fact that row price indexes are available makes the work easier. Despite being true that no major changes in prices have occur within the time period considered, the correspondent adjustments have to be done. Input-output demand models need to be fully calibrated, otherwise they become inconsistent. A file that shows how tables were transformed into constant prices is attached. In addition, the first author has designed the Path-RAS methodology (included in the 2018 UN Handbook on Supply and Use Tables and Input-Output Tables). This algorithm is able to balance matrices in limited information contexts. It could be used to adjust these tables in a context were only some column information is available.

Comment 3.

Normalizations are included in order to see the effects of the own consumptions made by each industry on its products. This is specified in the article several times. We understand that this way some technological changes could be questioned and other ratified with a higher level of detail.

Comments 4, 5 and 6.

Firstly, a change on the title has been decided. This way, we want to focus on the methodological proposal rather than the specific sectors included in its empirical application. We use these three sectors just for efficiently visualizing an empirical application. In the attached file it can be seen how calculations were done for all industries. It would be difficult to graphically show all results at the same time. We propose to publish a supplementary data file alongside the present article.

In second place, we think this decomposition methodology is compatible with others present in literature. The relations between them could even be formulated in mathematical terms. Another item to be considered would be to find an adequate visualization technique for the empirical results.

Last, a new reference were the two authors participated has been included in the conclusions section.

Round 2

Reviewer 1 Report

I think the authors have sufficiently responded to my comments and the revised manuscript is suitable for publication.

Author Response

The authors would like to thank the referee for his/her valuable comments during the revision process. 

Reviewer 3 Report

The paper has no major mistakes and it is much better than the last version.  It still presents a very simple form of structural decomposition analysis and in that sense is not innovative. I don't think that the option to use this example simply in primary sectors and in European countries for a short time period makes sense. The authors say that they do that because there are not important changes in prices. However, there are methods available to transform the table from current prices to constant and the reference to those methods should be included. 

Author Response

In order to perform SDA, price deflation is needed. Here, matrices were deflated considering 2010 prices as base year. For clarification, the following bibliographical reference is inserted:

Dietzenbacher, Erik and Alex R. Hoen 1998. Deflation of input-output tables from the user’s point of view: A heuristic approach. Review of Income and Wealth 44: 111–122.